How AWS Empowers 40 Startups to Scale Gen AI Technology

The race to spot the most promising AI startups is accelerating as the challenges of understanding, deploying and monetising AI deepen.

Leading cloud providers such as Amazon, Microsoft and Google are pouring millions into cloud credits and technical mentoring for early-stage ventures – betting that today’s innovators could become tomorrow’s enterprise heavyweights.

For smaller AI players, access to computing power often defines whether they scale efficiently or drain precious venture capital on infrastructure costs.

Now, AWS has revealed the latest cohort selected for its third Generative AI Accelerator programme, granting each participant up to US$1m in cloud computing credits.

The eight-week initiative gathers 40 startups developing solutions ranging from Arabic language models to molecular design systems driving the next wave of drug discovery.

What does AWS’ selection reflect about the AI market?

The selection underscores how AI innovation has fragmented into a multitude of specialised domains.

Some companies are addressing linguistic gaps where mainstream models struggle to perform.

Trillion Labs is training models tailored to Korean users, SCB 10X’s Typhoon project is advancing AI for Thai and Lisan AI is empowering Arabic-speaking professionals across government and business.

These projects are vital because most large language models (LLMs) remain predominantly trained on English data—limiting their true global utility.

“Whether it’s in biotech labs, creative studios or industrial applications, the pace of Gen AI innovation is extraordinary – and it’s happening everywhere,” says Sherry Karamdashti, General Manager (GM) and Head of Startups in North America at AWS.

Some of the startups advancements in drug discovery and financial automation 

Healthcare organisations are increasingly applying AI to solve targeted challenges in drug development rather than pursuing broad, exploratory research – and startups are playing a pivotal role in this shift.

Chai Discovery, for instance, is training models to design and optimise new molecules, while Manifold Bio integrates AI-driven protein engineering with real-world testing in living organisms.

Meanwhile, SyntheticGestalt has developed what it calls a molecular-focused foundation model – though such ambitious claims are not unusual in a sector where emerging companies often outpace their own capacity for delivery.

In the financial sector, standout selections include Hyperbots, which has created an agentic AI platform designed for finance teams.

Unlike conventional systems that simply respond to queries, agentic AI can take autonomous actions.

Hyperbots’ HyperLM is a language model trained specifically on financial data, giving it domain-specific precision.

Eloquent AI is pursuing similar automation for regulated processes, while Synthera AI is developing advanced tools for fixed-income modelling.

How robotics startups are targeting manufacturing gaps

The robotics cohort highlights how AI is advancing into physical domains that have long resisted full automation.

RLWRLD is building foundation models for industrial robots, training them on what it describes as high-precision motion data.

Mimic Robotics is developing systems tailored for retail and manufacturing environments, while Basetwo AI offers platforms that analyse data from pharmaceutical plants to recommend actions for engineers.

At the same time, a new wave of infrastructure startups is tackling the escalating costs associated with operating large-scale AI systems.

Inception Labs claims its Mercury system runs up to 10 times faster and cheaper than existing language models, leveraging what it describes as a diffusion-based approach.

Inephany, meanwhile, develops optimisation tools that enable companies to train AI models more efficiently – an essential capability when a single training cycle can cost hundreds of thousands of dollars.

Each participating startup receives technical and business mentoring alongside the awarded credits.

Through its accelerator, AWS provides guidance on machine learning performance, infrastructure configuration, and go-to-market strategy.

All participating founders will showcase their progress at AWS re:Invent in Las Vegas this December, connecting directly with potential investors and enterprise customers.

“This year’s cohort reinforces our mission to help that innovation move faster and deliver real-world impact for customers in every industry,” Sherry says.

“We’re removing the barriers and accelerating opportunities so these leaders can grow their world-changing solutions.”